Update app.py

app.py

@@ -26,13 +26,12 @@ SYSTEM_PROMPT = (
     "You are Talk GTE — a friendly AI assistant created by Vibow AI. "
     "GTE means Generative Text Expert in Vibow AI. "
     "Vibow AI created in 29 June 2025 and Talk GTE created in 23 October 2025. "
-    "Talk GTE have ≈ 1 trillion params. "
-   "Max tokens of Talk GTE can different, maybe max tokens of Talk GTE have ≈ 3600 tokens. "
+    "Talk GTE have approximately 1 trillion parameters. "
     "Stay positive, kind, and expert. "
     "Always capitalize the first letter of sentences. "
     "If the user requests code, always use triple backticks (```). "
-    "Be concise, neutral, and double-check your answers. "
-    "Sometimes, use emoji but relevant. "
+    "Be concise, neutral, and accurate. "
+    "Sometimes use emoji but relevant."
 )
 
 # =========================
@@ -56,7 +55,7 @@ def transcribe_audio(file_path: str) -> str:
             os.remove(file_path)
 
 # =========================
-# 🗣️ TTS
+# 🔊 TTS
 # =========================
 def text_to_speech(text: str) -> bytes:
     try:
@@ -72,9 +71,9 @@ def text_to_speech(text: str) -> bytes:
         return b""
 
 # =========================
-# 🌐 SerpAPI Search (now with images)
+# 🌐 SerpAPI with 3 images
 # =========================
-def serpapi_search(query: str, location: str = None, num_results: int = 3):
+def serpapi_search(query: str, location=None, num_results=3):
     url = "https://serpapi.com/search.json"
     params = {
         "q": query,
@@ -91,7 +90,6 @@ def serpapi_search(query: str, location: str = None, num_results: int = 3):
         results = []
         images = []
 
-        # text results
         if "organic_results" in data:
             for item in data["organic_results"][:num_results]:
                 results.append({
@@ -100,7 +98,6 @@ def serpapi_search(query: str, location: str = None, num_results: int = 3):
                     "link": item.get("link", "")
                 })
 
-        # image results
         if "images_results" in data:
             for img in data["images_results"][:3]:
                 images.append(img.get("thumbnail", ""))
@@ -112,17 +109,16 @@ def serpapi_search(query: str, location: str = None, num_results: int = 3):
         return {"results": [], "images": []}
 
 # =========================
-# 💬 Chat Stream
+# 💬 Stream Chat
 # =========================
 def stream_chat(prompt: str, history=None):
     wib = timezone(timedelta(hours=7))
     now = datetime.now(wib)
-    formatted_time = now.strftime("%A, %d %B %Y — %H:%M:%S WIB")
-
-    sys_prompt = SYSTEM_PROMPT + f"\nCurrent date and time: {formatted_time}."
+    sys_prompt = SYSTEM_PROMPT + f"\nCurrent time: {now.strftime('%A, %d %B %Y — %H:%M:%S WIB')}."
 
     messages = [{"role": "system", "content": sys_prompt}]
-    if history: messages += history
+    if history:
+        messages += history
     messages.append({"role": "user", "content": prompt})
 
     payload = {
@@ -133,61 +129,79 @@ def stream_chat(prompt: str, history=None):
         "stream": True,
     }
 
-    headers = {"Authorization": f"Bearer {GROQ_API_KEY_1}", "Content-Type": "application/json"}
-
-    try:
-        with requests.post(GROQ_URL_CHAT, headers=headers, json=payload, stream=True) as r:
-            for line in r.iter_lines():
-                if not line: continue
-                line = line.decode("utf-8")
-                if line.startswith("data: "):
-                    data = line[6:]
-                    if data == "[DONE]": break
-                    try:
-                        delta = json.loads(data)["choices"][0]["delta"].get("content", "")
-                        if delta: yield delta
-                    except: continue
-    except Exception as e:
-        print(f"[Stream Error] {e}")
+    headers = {"Authorization": f"Bearer {GROQ_API_KEY_1}"}
+
+    with requests.post(GROQ_URL_CHAT, headers=headers, json=payload, stream=True) as r:
+        for line in r.iter_lines():
+            if not line:
+                continue
+            line = line.decode()
+            if line.startswith("data: "):
+                data = line[6:]
+                if data == "[DONE]":
+                    break
+                try:
+                    delta = json.loads(data)["choices"][0]["delta"].get("content", "")
+                    if delta:
+                        yield delta
+                except:
+                    continue
 
 # =========================
-# 🧠 /chat endpoint
+# 🚀 Chat Endpoint (Text + Voice)
 # =========================
 @app.route("/chat", methods=["POST"])
 def chat():
+    # 🎤 Voice Mode
+    if "audio" in request.files:
+        audio = request.files["audio"]
+        temp = f"/tmp/{time.time()}_{random.randint(1000,9999)}.wav"
+        audio.save(temp)
+
+        user_text = transcribe_audio(temp)
+        if not user_text:
+            return jsonify({"error": "Failed STT"}), 500
+
+        keywords = ["hotel", "mall", "resort", "villa", "tempat wisata"]
+
+        if any(k in user_text.lower() for k in keywords):
+            serp = serpapi_search(user_text)
+            text = "\n".join([f"{r['title']} — {r['snippet']} — {r['link']}" for r in serp["results"]])
+            imgs = "\n".join(serp["images"])
+            user_text = f"{user_text}\n\nGoogle Results:\n{text}\n\nImages:\n{imgs}\n\nExplain & recommend."
+
+        ai = "".join(chunk for chunk in stream_chat(user_text))
+        audio_bytes = text_to_speech(ai)
+
+        return jsonify({
+            "mode": "voice",
+            "transcript": user_text,
+            "reply_text": ai,
+            "audio_base64": "data:audio/mp3;base64," + base64.b64encode(audio_bytes).decode()
+        })
+
+    # 💬 Text Mode
     data = request.get_json(force=True)
-    prompt = data.get("prompt", "").strip()
+    prompt = data.get("prompt", "")
     history = data.get("history", [])
-    if not prompt: return jsonify({"error": "No input text provided"}), 400
 
-    keywords = ["hotel", "mall", "resort", "tempat wisata", "villa"]
+    keywords = ["hotel", "mall", "resort", "villa", "tempat wisata"]
 
-    # add search results if needed
     if any(k in prompt.lower() for k in keywords):
         serp = serpapi_search(prompt)
         text = "\n".join([f"{r['title']} — {r['snippet']} — {r['link']}" for r in serp["results"]])
         imgs = "\n".join(serp["images"])
-
-        combined = (
-            prompt
-            + "\n\nGoogle Results:\n"
-            + text
-            + "\n\nImages:\n"
-            + imgs
-            + "\n\nExplain & recommend."
-        )
-    else:
-        combined = prompt
+        prompt = f"{prompt}\n\nGoogle Results:\n{text}\n\nImages:\n{imgs}\n\nExplain & recommend."
 
     def generate():
-        for chunk in stream_chat(combined, history):
+        for chunk in stream_chat(prompt, history):
             yield chunk
 
     return Response(generate(), mimetype="text/plain")
 
 # =========================
-# 🚀 Run
+# ▶️ Run
 # =========================
 if __name__ == "__main__":
-    print("🚀 Vibow AI Chat Server started")
+    print("🚀 Vibow Talk GTE Server Running")
     app.run(host="0.0.0.0", port=7860, debug=True, threaded=True)